ATAC tutorial

[revolvefire]

Congratulations on the wonderful-looking update on VIA.

This is my first time trying to use VIA (or StaVia), and it seems like the tutorial for [scATAC-seq Hematopoiesis] link is not working correctly yet. Will it be available soon? I am mostly interested in scATAC-based trajectory analysis and also wonder whether typically preprocessed scATAC data using Signac or ArchR would be applicable.

Thanks!

[ShobiStassen]

Hi! Thanks so much for reaching out. Let me check and get back to you in a couple of days. We are in the process of updating the tutorials to reflect the changes in the newer versions!

…

On Thu, 15 Feb 2024, 12:35 revolvefire, ***@***.***> wrote: Congratulations on the wonderful-looking update on VIA. This is my first time trying to use VIA (or StaVia), and it seems like the tutorial for [scATAC-seq Hematopoiesis] link is not working correctly yet. Will it be available soon? I am mostly interested in scATAC-based trajectory analysis and also wonder whether typically preprocessed scATAC data using Signac or ArchR would be applicable. Thanks! — Reply to this email directly, view it on GitHub <#57>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AISI4SHGZ2FLTL3ZD5F5INDYTWGC7AVCNFSM6AAAAABDJSE2OSVHI2DSMVQWIX3LMV43ASLTON2WKOZSGEZTKNRRHA3DQNI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[ShobiStassen]

@revolvefire
Hi, i've put up a simple scATAC-seq tutorial for you, hopefully it helps get you started

[revolvefire]

@ShobiStassen

Thank you so much for the prompt update! I'll start looking into it!
Thank you!

[A-legac45]

@ShobiStassen @revolvefire

Hi, I am interested by trying to use your tools with the goal to realize trajectory which will combine scmultiomic data (snatac-se and snRNA-seq) inforlations? I am wondering If this tools can work with both information? How will it consider both information when performing the trajectory analysis? the other tools such as slingshot ... is not able to manage this. Which type of files do you need to start with only snRNA assay or can it take also in account at the same time the peak matrix information? Thanks in advance

[ShobiStassen]

@A-legac45
Hi, generally speaking you would need to integrate your data and then run some for of dimensionality reduction (like PCA, typically 10-100PCs) before passing your features to StaVia

[A-legac45]

@ShobiStassen Hi, I do not need to integrate my data with 10x mutiom technology I have both information directly for one nuclei. I am more concerned about the clustering I need to process it before based on both data by using a LSI? Does somebody already performed stavia on this type of data? What is the advantage to use stavia compare to monocle or slingshot? Do you have a tutorial? Thanks for your answers. Bests

[ShobiStassen]

hi @A-legac45 , Im not too familiar with the details of a joint analysis for paired data where the cells are the anchor points (i.e. you have multimodal measurements for each cell). you could consider getting a low dimensional representation from MOFA+ (Argelaguet, R. et al. MOFA+). This article talks more about various options for pre-processing scatac+scrnaseq data so you get a shared low-dimesional representation. This is probably useful for visualiation, but I'm not sure if these joint low-dim representations result in a lot of information loss compared to handling each type of data individually.

If you want, it is also perhaps more straightforward first to handle each type of measurement separately and then compare the resulting analyses. i.e. one analysis for the rna-seq (do HVG followed by PCA before StaVia) and one analysis for the atac-seq (using LSI prior to using StaVia for clustering and TI) and then look at the trajectories in each case for similarities and differences. there are tutorials on the RTD pages.

for a comparison to Monocle or Slingshot, we find that StaVia gives more control to the user in terms of the desired detail of lineages and trajectories, clustering, or single-cell level views and also lots of visualizations - you can look at the latest paper for some examples of analyses.
Our first paper on Via 1.0 has some discussion of the advantages compared to slingshot and monocle specifically, but refers to the older version of StaVia

[A-legac45]

Hi @ShobiStassen thanks for your answer, so I f I get it properly for stavia I will need to give a star and end cluster for the trajectories it will not predict the trajectory from a starting point ? So If I think I have bifurcation clusters I need to know previously where does it occur? bests

[ShobiStassen]

hi @A-legac45
you only need to provide a start state. the ends states are auto-predicted.
the start state can be auto-predicted if you have RNA-velocity, but I would say that if you have an idea for a suitable start state, you can use that first. You can always increase the resolution of clustering to narrow down the location of bifurcation